Automatic classification of canine PRG neuronal discharge patterns using K-means clustering

Area of Science:

• Neuroscience
• Computational Biology
• Respiratory Physiology

Background:

• Respiratory-related neurons in the parabrachial-Kölliker-Fuse (PB-KF) region are crucial for breathing control.
• Pontine respiratory group (PRG) neurons display diverse discharge patterns (inspiratory, expiratory, phase-spanning, non-respiratory).
• Classifying these diverse neuronal patterns into distinct subgroups based on discharge contours is challenging.

Purpose of the Study:

• To present an automated method for classifying neuronal discharge patterns.
• To derive average discharge contours for identified neuron subgroups.
• To objectively categorize PRG neuron activity and identify archetypal patterns.

Main Methods:

• Utilized K-means clustering technique for automated classification of neuronal discharge patterns.
• Implemented the method using SigmaPlot User-Defined transform scripts.
• Classified discharge patterns of 135 canine PRG neurons and described methods for optimal cluster number selection.

Main Results:

• Successfully classified the discharge patterns of 135 canine PRG neurons into seven distinct subgroups.
• The K-means clustering method provided a robust and objective means of categorization.
• Identified archetypal contours representing subtypes based on discharge patterns.

Conclusions:

• The K-means clustering method offers an objective approach to automatically categorize neuron discharge patterns.
• This technique facilitates the identification of underlying archetypal contours for neuronal subtypes.
• The developed method aids in understanding the functional diversity of PRG neurons.